Rhodes is Careening Down the Wrong Road
By Rowell Huesmann and Leonard Eron
The University of Michigan

It is not surprising that Jack Valenti and his employers who make vast amounts of money off of violence should attack the messengers who tell them that they are harming our kids. Nor is it surprising that those whom William Bennett has called the "paid shills of the entertainment industry" would once again be trying to deny that media violence has any effect on a child's behavior. What is surprising is that a science writer who seems to be as well informed as Richard Rhodes could also blind himself to the truth about the issue. 

In his New York Times Op-Ed article (September 17, 2000) and his Rolling Stone piece in the November 23 issue, Mr. Rhodes blends small facts with large fictions to paint a picture that media violence has no effect worth talking about. The truth is quite different. Twenty-eight years ago in 1972 the Surgeon General of the United States issued a warning that media violence is harmful to some youth. Eighteen years ago the National Institutes of Mental Health issued a report reinforcing that conclusion. The recent statements of the AMA and other professional organizations are only the latest in a long stream of reports reaching the same conclusion. And in January 2000 the Surgeon General released a report on youth violence that concluded that exposure to media violence is one important risk factor for youth violence about which we should be concerned.

Are these reports at odds with the opinions of most researchers working in the area as Mr. Rhodes suggests? No! Over 80% of the researchers working in the area concur with these reports. They believe that the 50 years of experimental studies on children and longitudinal studies following children as they grow up have painted a conclusive picture indicting media violence as one contributing cause of youth and young adult violence.

But let us consider some of the specific claims Mr. Rhodes makes in his inflated rhetoric.

Do the well-controlled laboratory experiments where children are randomly assigned to watch a violent film or a non-violent film show no effect about as often as they show an effect as Mr. Rhodes implies? No! These laboratory experiments consistently show a strong effect for media violence on aggressive behavior. In the best meta-analytic review of recent years, George Comstock and Haejung Paik analyzed 217 key studies and reported that in 586 comparisons of randomly assigned groups, the "average effect size" was .42. This means that if the probability of behaving violently before watching the film were about 50/50, watching the film on the average increased the probability to about 70/30. This is a huge effect size by the standards of social science research. Furthermore, despite Mr. Rhodes assertion to the contrary, these studies do show that media violence is causing aggressive behavior. As Mr. Rhodes should know, causation can only be demonstrated by conducting well-controlled trials in which children are assigned randomly to see one film or another. That is what these experiments do.

Do these controlled laboratory and field experiments deal mostly with verbal or simulated aggression as Mr. Rhodes suggests. No! Comstock and Paik identified 71 comparisons in which the criterion behavior was actual physical violence against another person. The average effect size was .33 for these comparisons, another very large effect. In one study normal nursery school children who had watched violence pushed and shoved their peers significantly more than those who had not. In another study adolescent boys who had watched violence got into more fights afterwards while playing a sport than those who had not. In still another study teen-aged boys in a home for delinquents who had watched violence got into more serious fights that week than those who had not watched violence.

Do the longitudinal field studies that have been done on the topic not add anything to the case for an effect as Mr. Rhodes asserts? No, they add a lot. In our own Columbia County, NY study we found that boys who watched a lot of violent television when growing up in the 60s became more aggressive and violent young men at age 19 in the 70s. Mr. Rhodes suggests that this result provides weak evidence because an effect was shown only on aggressiveness as rated by peers and not on self-reports or personality tests. However, almost all researchers working in this area consider peer-rated aggressive behavior to be the best measure one can use to assess youth violence. Adolescents often lie when reporting on their own behavior, and a personality test is not a measure of how one actually behaves. Thus the validity of self-reports and personality tests is questionable but not so for peer reports.. Another important point not mentioned by Mr. Rhodes is that exposure to media violence at ages 6 to 11 predicts increased aggression in young adulthood for those kids who are initially low on aggression as well as those initially high on aggression, for kids of poor parents and wealthy parents, for low IQ kids and high IQ kids. Furthermore, there are now other longitudinal studies which reinforce these points.

In another study that tracked Children in Oak Park, Illinois from age 6 to about age 25, we found that both the boys and girls who habitually watched more TV violence when they were 6 to 10 years old grew up to be more aggressive in their 20s. What do we mean by more aggressive in their 20s? They reported engaging in more physical aggression such as shoving, hitting, and choking others. Friends, who knew them well reported the same about them. Even when one statistically equated them for aggressiveness in childhood, one found that the children who watched more TV violence increased more in aggressiveness. The high violence viewing boys and girls both also grew up to abuse their spouses more. The high violence viewing boys were also found to have more criminal convictions in the Illinois records and more moving traffic violations. Mr. Rhodes does not mention any of this in his inflammatory articles. Perhaps he did not know about these results. No, he knew about them because we long ago told him about these results.

Instead of objectively evaluating all this evidence Mr. Rhodes prefers to resort to ad hominem attacks and to accuse us of fraud. He picks out one result that we had presented in the mid-1980s from the Columbia County study and asserts that it represents fraud. He maintains that we concluded that early TV violence viewing causes adult criminal behavior on the basis of data about only 3 boys. Mr. Rhodes has his statistics mixed up. Here are the facts. There were 145 boys in that study on whom we had data about how violent was the TV they watched at age 8 and how aggressive and criminal they were at age30. We found a statistically significant correlation of about .20 between the violence those 145 boys were watching on TV at age 8 and their self-reported aggressiveness at age 30. Such an effect size is similar to what has been found for many risks to public health that have generated universal concern, e.g. children's IQ scores and exposure to lead, lung cancer and passive exposure to cigarette smoke. When we looked at serious violent crimes we found that 3 of the 145 boys had been arrested for such crimes. This is about typical in this country -- not many people are arrested for serious violent crimes. To our great concern we found that all three of those boys who had been arrested for serious violent crimes were in the highest 40% of childhood violence viewers and 2 were in the highest 20%. Mr. Rhodes disparages these results as a fraud. Why are they a fraud? We never hid the exact results. We urged people to treat them with caution until they were confirmed by other results. They have been confirmed by other results. We think Mr. Rhodes is blinded by his own biases as we discuss below.

What makes the case against TV violence so strong is the convergence of the randomized experiments demonstrating causation and the longitudinal studies like the ones we have done demonstrating lasting effects. This convergence should bolster any unbiased person's confidence in the truth of the conclusion that media violence increases aggression. 

Given these facts, we are puzzled by Mr. Rhodes reluctance to accept the conclusion that media violence stimulates aggression in children, and relates to future aggression. However, we must acknowledge that Mr. Rhodes is not alone in his skepticism. 

Part of the problem may be that science never produces uniform results. There are always some number of studies that show no effect. There are always some researchers who disagree. However, the fact is that research results are even more ambiguous in areas that Mr. Rhodes and others accept uncritically. For example, Mr. Rhodes asserts that "there is good evidence, causal evidence ... that (violence) is learned in personal violent encounters, beginning with the brutalization of children by their parents or their peers." Well, Mr. Rhodes should know better. There is NO scientific causal evidence at all to this effect, only anecdotal evidence. There cannot be causal evidence because it is impossible ethically to do randomized experiments on this topic. Anyone who randomly assigned children to experience personal violent encounters would quickly be arrested. This is not to say that Mr. Rhodes is wrong about personal violent encounters causing violence. But there are not and never will be studies that actually demonstrate that fact. Similarly, while most of us accept the fact that smoking causes lung cancer, there are no studies that demonstrate that causation because one cannot ethically randomly assign one group of persons to smoke cigarettes. In fact, the correlational evidence implicating smoking as a risk factor for cancer is not any stronger than the correlational evidence implicating media violence as a risk factor for violent behavior.

Nowhere have we ever indicated that media violence is the only or even a major cause of violence among youth. All indications from the meta-analyses mentioned above are that television can account for 10% of youth violence. This means that 90% is caused by other factors. However, it would be no mean achievement if we could lower the frequency of violence in this country by 10%. Physical violence within the home and neighborhood may certainly be another factor. Mr. Rhodes believes that this is the major, if not only cause. However, he produces no evidence for this based on any carefully designed scientific study. It is only his opinion bolstered perhaps by the statements of others who also present no hard evidence. Moreover, his personal experience would seem to contradict this theory as much as his viewing TV violence and not behaving aggressively contradicts that theory. According to his own report, he was raised in a violently abusive home. Yet, as far as we know, he has not grown up to be a violent person unless you consider verbal attack and misrepresentation as an indication of violence.

Another contributor to the problem may be the visceral reaction that anything that hints at censorship produces in writers and artists. Mr. Rhodes knows very well that there will never be any real censorship of the mass media in this country, but that may not prevent the emotional affect that the idea generates. However, we suspect that for many people like Mr. Rhodes who are genuinely concerned about violence in this country, the most important reason for their reluctance to accept media violence as a cause may be their fear that another cause of violence that they consider even more important may be ignored. This could be the case if we allow it to happen, but Mr. Rhodes is wrong if he thinks it is already the case that a focus on media violence has hurt funding for his personal concern -- the building of community mental health centers. The total amount of money spent on research on media violence by NIH in the past five years is in fact quite small. It has hardly had an impact on funding for research in any other area. Furthermore, the very longitudinal study that we conducted and which does not impress Mr. Rhodes, was conducted as part of the activities of one of the very first community mental health centers in this country -- The Rip Van Winkle Clinic in Columbia County, New York.

It is also clear that Mr. Rhodes is not an unbiased observer. He has a conflict of interest when he writes on violence. Not only is his own ego wrapped up in the view that he has been damaged by being abused as a child, but his own financial well being depends on the sales of his book, "Why They Kill" which takes the viewpoint that media violence is unimportant and being violently abused is important. This is a questionable thesis as we indicated above and as the Surgeon General indicated in his January report on youth violence. In that report he lists seven great myths about violence, and one of them is that child abuse inevitably leads to violence later in life. In the past Mr. Rhodes has been known to review negatively books that competed with his own books (see his review of Philip Caputo's "Horn of Africa") when other more ethically concerned writers might have declined to review a competing book; so it is hard to see his comments on this topic as those of an unbiased observer. Finally, if he were really an unbiased observer, he might discuss the evidence that individual differences in testosterone plays a role in aggression. But as one who has publicly confessed to taking testosterone regularly, that might have been hard for him to do.

Finally, it seems likely that skeptics like Mr. Rhodes are victims of a variation of the so-called "third person effect." We all tend to think that many risks in life may be risks for "other people," but not "for me and my family." Most of us in Mr. Rhodes generation grew up watching violent films and TV, and most of us do not behave very aggressively. Therefore, it is hard for many of us to believe that media violence can be having an effect on us. But be assured. Just as every cigarette your child smokes raises the odds that some day your child may get lung cancer, every violent scene your child sees raises the odds that some day your child may behave more violently than they otherwise would.